US9646266B2 - Feature type spectrum technique - Google Patents
Feature type spectrum technique Download PDFInfo
- Publication number
- US9646266B2 US9646266B2 US15/142,099 US201615142099A US9646266B2 US 9646266 B2 US9646266 B2 US 9646266B2 US 201615142099 A US201615142099 A US 201615142099A US 9646266 B2 US9646266 B2 US 9646266B2
- Authority
- US
- United States
- Prior art keywords
- feature
- features
- frequencies
- textual input
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G06N99/005—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G06F17/28—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/40—Software arrangements specially adapted for pattern recognition, e.g. user interfaces or toolboxes therefor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/40—Processing or translation of natural language
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/20—Drawing from basic elements, e.g. lines or circles
- G06T11/206—Drawing of charts or graphs
Definitions
- Design fixation is the tendency to fixate on the features of known solutions when trying to create novel solutions (Jansson & Smith, 1991). For example, a subject who is shown an existing chair and then asked to design an improved chair is likely to fixate on features of the existing chair when attempting to design an improved chair. Such fixation can lead the subject to overlook features that would be useful to include in an improved chair, but which are lacking in the existing chair.
- Sensors are used to generate sample set data representing objects in a sample set.
- a computer system analyzes the sample set data to determine the frequencies with which features in a feature set are observed in the objects in the sample set.
- An example of such output is a bar chart representing the frequency of observation of features in the feature set in a particular object.
- the feature output may be used to identify one or more obscure (i.e., low frequency) features in the particular object.
- Machine learning may be used to learn associations between sample set data and features in the feature set, thereby improving the accuracy and efficiency of future uses of the computer system.
- FIG. 1 is a bar chart representing an example of feature output according to one embodiment of the present invention
- FIG. 2 is an illustration of a part of a feature set, also referred to herein as a feature type taxonomy, according to one embodiment of the present invention
- FIG. 3 is an illustration of a plastic chair
- FIG. 4 is a dataflow diagram of a system for assisting in overcoming design fixation according to one embodiment of the present invention
- FIG. 5 is a flowchart of a method performed by the system of FIG. 4 according to one embodiment of the present invention.
- FIG. 6 is a flowchart of a method performed by the system of FIG. 4 to use machine learning to learn associations between sample set data and features according to one embodiment of the present invention.
- FIG. 7 is a dataflow diagram of a system for implementing the method of FIG. 6 according to one embodiment of the present invention.
- Embodiments of the present invention may be used to alleviate design fixation in a variety of ways.
- a dataflow diagram is shown of a system 400 that may be used to alleviate design fixation according to one embodiment of the present invention.
- FIG. 5 a flowchart is shown of a method 500 performed by the system 400 of FIG. 4 according to one embodiment of the present invention.
- sample set For a set of objects in a particular class of objects, such as a set of chairs in the class of chairs. Such a set of objects in a particular class of objects will be referred to herein as a “sample set.”
- the system 400 of FIG. 4 includes sample set data 402 representing the sample set.
- the sample set data 402 may, for example, be computer-readable data representing the objects in the sample set.
- the sample set data 402 may be data stored in a non-transitory computer-readable medium.
- the sample set data 402 may, for example, be in the form of a database that includes one record for each of the objects in the sample set.
- Data representing an object in the sample set may take any form in the sample set data 402 , such as a digital image of the object, a two-dimensional or three-dimensional model of the object, a textual description of the object, a parameterized model of the object (containing one or more parameters and corresponding parameter values), or any combination thereof. These are merely examples, however, and do not constitute limitations of the present invention.
- the sample set data 402 may take any form consistent with the description herein.
- the sample set may include any number of objects.
- the sample set may consist of a single object.
- the sample set may, however, include two, three, or more objects, without any limit.
- the sample set data 402 may represent solely a single object, or two, three, or more objects, without any limit.
- the objects in the sample set may have features that differ from each other.
- one chair in the sample set may have four legs while another chair in the sample set may have three legs.
- one chair in the sample set may be constructed from plastic while another chair in the sample set may be constructed from wood.
- one object in the sample set may be a rocking chair, which is capable of moving during its normal course of use, while another object in the sample set may be a conventional dining room chair, which is stationary during its normal course of use.
- One function that may be performed by the system 400 of FIG. 4 and the method 500 of FIG. 5 is to identify features of the objects in the sample set.
- the system 400 may include a feature identification module 406 a , which may identify features of the objects in the sample set based on the sample set data 402 , thereby producing feature data representing the identified features of the sample set.
- Embodiments of the present invention may use a feature set, also referred to herein as a “feature type taxonomy.”
- the feature set may include any number of features, examples of which will be described below.
- the system 400 may include feature set data 404 , which may represent the feature set.
- the feature set data 404 may, for example, be computer-readable data representing the features in the feature set.
- the feature set data 404 may be data stored in a non-transitory computer-readable medium.
- the feature set data 404 may, for example, be in the form of a database that includes one record for each of the features in the feature set.
- Data representing a feature in the feature set may take any form in the feature set data 404 , such as a textual name of the feature, a definition of the feature, a human-readable description of the feature, or any combination thereof. These are merely examples, however, and do not constitute limitations of the present invention. In general, the feature set data 404 may take any form consistent with the description herein.
- the system 400 may determine whether each object in the sample set has each of the features in the feature set.
- the system 400 may, for example, make such determinations based on the sample set data 402 and/or the feature set data 404 .
- the result of such a determination for each feature-object pair may, for example, be a binary value (representing, e.g., “has” or “does not have”) for that feature-object pair.
- This set of binary values (one for each feature-object pair) may be contained within the feature data 408 that is output by the feature identification module 406 a.
- the feature identification module 406 a may include: (1) one or more computers; (2) one or more humans; or (3) any combination of (1) and (2).
- the feature identification module 406 a may include a computer that automatically generates and/or analyzes some or all of the sample set data 402 to produce some or all of the feature data 408 based on some or all of the feature set data 404 .
- the feature identification module 406 a may include a human who manually analyzes some or all of the sample set data 402 to produce some or all of the feature data 408 based on some or all of the feature set data 404 .
- the functions performed by the feature identification module 406 a may be divided between computers and humans in any of a variety of ways.
- a computer may produce feature data 408 for one object represented by the sample set data 402 automatically, while a human may produce feature data 408 for another object represented by the sample set data 402 manually.
- a computer may produce feature data for certain features of an object automatically, while a human may produce feature data for other features of the same object automatically, in which case the feature data 408 produced for that object will include some feature data produced by the computer and other feature data produced by the human.
- the human may directly observe objects in the sample set using the human's senses, such as by looking at the object, touching the object, listening to the object, smelling the object, tasting the object, or any combination thereof.
- the sample set data 402 may include the objects in the sample set themselves, either in addition to or instead of data representing the objects in the sample set.
- the human may produce some or all of the feature data 408 based on digital sample set data 402 , such as digital images of the objects in the sample set, or on other indirect input containing information about the objects in the sample set, rather than based on direct sensory perception of those objects.
- the system 400 may use the feature data 408 to produce feature output 416 representing the features of the sample set represented by the feature data 408 .
- the feature output 416 may, for example, represent the frequency of occurrence of each feature in the feature data 408 .
- the feature output 416 may indicate the number of occurrences of the motion feature in the feature data 408 .
- the feature output 416 may take any of a variety of forms, such as graphical output (e.g., a bar chart or other chart).
- the feature data 408 may be used to generate the feature output 416 in any of a variety of ways.
- the system 400 may include a feature count module 410 .
- the feature count module 410 may generate, based on the feature data 408 , for each feature in the feature set (represented by the feature set data 404 ), a count of the number of occurrences of the feature in the feature data 408 .
- the count of the number of occurrences of a feature in the feature data 408 is referred to herein as the feature's “frequency count.”
- the frequency count for a particular feature may be obtained, for example, by summing the binary values corresponding to the particular feature in the feature data 408 .
- the feature count module 410 may produce feature count data 412 , which may include frequency counts for some or all of the features in the feature set (represented by feature set data 404 ) and for some or all of the objects in the sample set (represented by the sample set data 402 ).
- the system 400 may include a feature count output module 414 , which may produce feature output 416 based on the feature count data 412 in any of a variety of ways.
- the feature count output module 414 may produce feature output 416 in the form of a chart, such as a bar chart, a pie chart, or other chart representing the frequency counts in the feature count data 412 . Because such a chart may resemble a spectrum of values, such a chart, or its underlying data, may be referred to herein as a “feature type spectrum.”
- a feature type spectrum may be understood not to be limited to any particular examples disclosed herein, such as bar charts, but instead to encompass any kind of output representing the feature count data 412 .
- the system 400 may include one or more additional feature identification modules, such as feature identification modules 406 b and 406 c .
- Each of the feature identification modules 406 a , 406 b , and 406 c may apply the techniques described above to the sample set data 402 and the feature set data 404 .
- the frequency counts produced by the feature identification modules 406 a - c may be aggregated (e.g., summed) with each other, so that the resulting feature data 408 represents the sums of the frequency counts produced by the feature identification modules 406 a - c.
- the feature identification module 406 a produces a frequency count of 1 for a particular feature of the sole object in the sample set
- that feature identification module 406 b produces a frequency count of 0 for the same feature of the sole object in the sample set
- that feature identification module 406 c produces a frequency count of 1 for the same feature of the sole object in the sample set
- the feature data 408 may include a value of two for the particular feature of the sole object in the sample set, as a result of summing 1, 1, and 0.
- the same technique may be applied to other features of the same object and to features of other objects (if the sample set contains other objects).
- the system 400 may include any number of feature identification modules, such as one, two, three or more feature identification modules.
- Each of the feature identification modules in the system 400 may be or include a computer, a human, or a combination thereof.
- each of the three feature identification modules 406 a - c may be a human.
- each of the three feature identification modules 406 a - c may be a computer.
- one of the feature identification modules 406 a - c may be a computer, while the other two of the feature identification modules 406 a - c may be humans.
- the method 500 illustrated by FIG. 5 is an example of a method that may be used to implement the techniques disclosed above.
- the method 500 may, for example, be performed in whole or in part by one or more of the feature identification modules 406 a - c .
- the method 500 begins by initializing the feature data 408 ( FIG. 5 , operation 502 ).
- the method 500 may, for example, initialize values corresponding to each of the features represented by the feature set data 404 to an initial value, such as zero.
- the method 500 enters a loop over each object O in the sample set represented by the sample set data 402 ( FIG. 5 , operation 504 ).
- the method 500 enters a loop over each feature F in the feature set represented by the feature set data 404 ( FIG. 5 , operation 506 ).
- the method 500 determines whether the object O has the feature F ( FIG. 5 , operation 508 ).
- the method 500 may make this determination in any of a variety of ways.
- operation 508 may be performed by: (1) receiving the sample set data 402 and the feature set data 404 as input; (2) observing, analyzing, or otherwise processing some or all of the sample set data 402 and some or all of the feature set data 404 to determine whether the object O has the feature F.
- the determination may, for example, be made by one of the feature identification modules 406 a - c . If the feature identification module that performs operation 508 is a computer, then the computer may make the determination using any of a variety of techniques.
- sample set data 402 may be pre-categorized by the creator of the sample set data 402 .
- sample set data for a cup might indicate explicitly that the cup is made of ceramic and that ceramic is a type of material, where material is a type of feature.
- a computer may determine that the cup is made of ceramic based directly on the data in the sample set data, without any further processing.
- the human may make the determination manually and provide input to the method using any suitable input device (such as a keyboard, mouse, microphone, touchscreen, or any combination thereof), wherein the input indicates whether the object O has the feature F.
- the system 400 and method 500 need not include the ability to determine whether object O has feature F automatically, but instead may rely on the judgment of the human, as represented by the input provided by the human to the system 400 and method 500 . If the human input indicates that the object O has feature F, then the method 500 concludes in operation 508 that the object O has feature F. Conversely, if the human input indicates that the object O does not have feature F (or if the human input does not indicate that the object O has feature F), then the method 500 concludes in operation 508 that the object O does not have feature F.
- the feature identification modules 406 a - c may receive some or all of the sample set data 402 from one or more devices, such as from one or more sensors. Any such sensor may perform a sensing operation on an object and generate output, within the sample set data 402 , representing a sensed property of the object. Any such output received from one or more sensors may be referred to herein as “sensor data.”
- the term “sensor” refers to a device, not to a human. Although a human may provide input to a sensor to cause the sensor to perform a sensing operation, the sensor nonetheless performs the sensing operation automatically, i.e., without human intervention. A sensor may be caused automatically to perform a sensing operation, i.e., not in response to input from a human. For example, a sensor may automatically perform sensing operations periodically, or in response to input from another device.
- the feature identification modules 406 a - c may receive sensor data automatically, i.e., without human intervention.
- the feature identification modules may receive a particular sensor datum automatically from a particular sensor by “pulling” that sensor datum automatically from the sensor (e.g., by sending a request automatically to the particular sensor for the sensor datum, and then receiving the sensor datum automatically from the particular sensor in response to the request), or by the particular sensor “pushing” that sensor datum automatically to one or more of the feature identification modules 406 a - c .
- Examples of such pushing include: (1) the sensor periodically (e.g., every second, minute, or hour) performing a sensing operation to generate sensor data and then automatically sending the sensor data to one or more of the feature identification modules 406 a - c ; and (2) the sensor detecting a change in the environment (e.g., the appearance of an object) and, in response to such detection, automatically performing a sensing operation to generate sensor data, and then automatically sending the sensor data to one or more of the feature identification modules 406 a - c.
- the sensor periodically (e.g., every second, minute, or hour) performing a sensing operation to generate sensor data and then automatically sending the sensor data to one or more of the feature identification modules 406 a - c .
- Embodiments of the present invention may use any of a variety of kinds of sensors to sense, and provide to the feature identification modules 406 a - c , sample set data including sensor data.
- sensors which may be used to perform sensing operations to generate, and send to the feature identification modules 406 a - c , sample set data including sensor data include any one or more of the following, in any combination:
- the method 500 stores a record (e.g., in the feature data 408 ) indicating that object O has feature F ( FIG. 5 , operation 510 ); otherwise, the method 500 stores a record (e.g., in the feature data 408 ) indicating that object O does not have feature F ( FIG. 5 , operation 512 ).
- operation 508 makes a binary determination of whether object O has feature F, resulting in a conclusion that object O either has or does not have feature F
- this is merely an example and does not constitute a limitation of the present invention.
- any feature may have one or more parameters, each of which may have a set of permissible values. For example, assume that feature F has parameters P 0 and P 1 , that parameter P 0 has a range of values V P0 (0) and V 0 (1), and that parameter P 1 has a range of values V P1 (0), V P1 (1), and V P1 (2).
- both objects O 0 and O 1 may have feature F, and both objects O 0 and O 1 may have parameter P 0 , but object O 0 may have a first value of parameter P 0 (such as value V P0 (0)), while object O 1 may have a second value of parameter P 0 (such as value V P0 (1)).
- Objects may have any number of parameters of a feature, and an object that has a particular parameter of a feature may have any value of that parameter.
- the feature of color may have a parameter of hue, which may have a range of values such as red, blue, and green.
- a parameter of hue which may have a range of values such as red, blue, and green.
- feature F is the feature of color
- one pen may have an ink color of blue
- another pen may have an ink color of green.
- Both pens have the feature of color and the parameter of hue, but each pen has a different value of that parameter.
- three plastic cups may all have the feature of size and the parameter of magnitude, but the first plastic cup may have a parameter value of small, the second plastic cup may have a parameter value of medium, and the third plastic cup may have a parameter value of large.
- An object may be said to “have” a parameterized feature if the object has any value of any parameter of that feature.
- an object may be said to have the feature of “color” if the object has any value of the “hue” parameter of color (e.g., red, blue, or green).
- An object may be said not to “have” a parameterized feature if the object does not have any value of any parameter of that feature (or if the object has a null value for the parameterized feature). For example, if the only parameter of the “color” feature is “hue,” and a particular object does not have any “hue” value (or has a null “hue” value), then the particular object may be said to lack the feature of “color.”
- Parameters and parameter values may be treated as features for any of the purposes described herein. For example, if the feature of “color” has parameters of “hue” and “intensity,” then the “hue” and “intensity” parameters may themselves be treated as features for any of the purposes described herein. For example, feature data 408 , feature count data 412 , feature output 416 , and obscure feature data 420 may be generated for parameters and parameter values. As a particular example, an object with a “hue” parameter value of “green” may be said to have the “hue” feature and the “green” feature (i.e., the feature of “green-ness”).
- Operation 508 may include correlating or mapping data, such as input provided by humans in the feature identification modules 406 a - c , to features, parameters, and parameter values. For example, one human observer may provide input describing a feature of a stapler as “staples paper,” while another human observer may provide input describing a feature of the same stapler as “fastens paper together.” Operation 508 may include determining that both such statements refer to the same feature and that both statements indicate that the stapler has that feature.
- FIG. 7 Examples of techniques for determining that both such statements refer to the same feature are illustrated by the method 600 of FIG. 6 and the system 700 of FIG. 7 .
- the system 700 of FIG. 7 may include all of the elements of FIG. 4 , only certain elements of FIG. 4 are shown in FIG. 7 for ease of illustration.
- FIG. 7 only shows feature identification module 406 a
- the system 700 of FIG. 7 may include additional feature identification modules 406 b - c .
- the feature identification module 406 a may map that first input to at least one first feature, parameter, and/or parameter value ( FIG.
- the feature identification module 406 a may map that second input to at least one second feature, parameter, and/or parameter value ( FIG. 6 , operation 608 ). Operations 602 , 604 , 606 , and 608 may be performed automatically by computer-implemented feature identification modules.
- the first input and the second input may take any form.
- both the first input and the second input may be or include textual input, e.g., text strings (such as “reduces vibrations” and “minimizes rattling”).
- the first input may be mapped to a first corresponding feature, parameter, or parameter value
- the second input may be mapped to a corresponding second feature, parameter, or parameter value. Solely for ease of explanation, the following description will refer solely to a feature, rather than a feature, parameter, or parameter value. However, it should be understood that any technique disclosed herein in connection with a feature is equally applicable to a parameter or parameter value.
- the feature identification module 406 a may determine whether the first feature is the same feature as the second feature ( FIG. 6 , operation 610 ). For example, the feature identification module 406 a may determine whether the first feature and the second feature are both the same feature in the feature set data 404 . In response to determining that the first feature and the second feature are the same feature, the feature identification module 406 a may determine that the first input and the second input both indicate that the object O has the feature F (which is both the first feature and the second feature) ( FIG. 6 , operation 612 ). As a particular example:
- the feature identification module 406 a may use any technique to determine whether the first input and the second input indicate the same feature as each other.
- the feature identification module 406 a may be computer-implemented and may use any combination of one or more digital thesauri, technical dictionaries, slang dictionaries, and urban dictionaries to determine that the first input and the second input indicate the same feature as each other.
- the feature identification module 406 a may look up the first input and the second input in a digital thesaurus and determine, based on the contents of the digital thesaurus, whether both the first input and the second input have the same meaning (e.g., both are synonyms for the same term). If the contents of the digital thesaurus indicate that the first input and the second input have the same meaning, then the feature identification module 406 a may conclude that the first input and the second input indicate the same feature as each other.
- the feature identification module 406 a may determine whether the first input and the second input indicate the same feature as each other based on mapping input 434 received from a user 436 .
- the mapping input 434 may indicate that the first input and the second input both indicate the same feature as each other.
- the mapping input 434 may, for example, contain data representing a particular feature that is indicated by both the first input and the second input. Additionally, the mapping input 434 may, for example, contain data representing or otherwise referring to the first input and/or the second input.
- the mapping input 434 may be input from the user 436 indicating that both the text (first input) “reduces vibrations” and the text (second input) “minimizes rattling” refer to the same feature of “motion.”
- the system 700 may store a mapping data structure 430 which contains a plurality of mappings 432 a - n of inputs to features, where n may be any number.
- the mappings 432 a - n alternatively or additionally, map inputs to parameters and/or parameter values.
- each of the mappings 432 a - n may represent a mapping between a particular input (e.g., text string) and a particular feature.
- one of the mappings 432 a - n may map the text string “fabric” to the feature of “material.”
- the feature identification module 406 a may map an input (such as the first input or the second input) to a feature by searching for the input in the mappings 432 a - n (e.g., using the input as an index into the mappings 432 a - n ) and, if a particular mapping containing the input is found in the mappings 432 a - n , then the feature identification module 406 a may identify the feature to which that input is mapped by the particular mapping. In this way the feature identification module 406 a may map the input to a corresponding feature.
- the feature identification module 406 a may store a record of this common mapping of the first input and the second input to the same common feature for future use ( FIG. 6 , operation 614 ). For example, in response to determining that the first input and the second input both indicate the same feature, the feature identification module 406 a may store, in the mappings 432 a - n:
- the feature identification module 406 a may store, in the mappings 432 a - n , a single mapping indicating that the first input and the second input map to the common feature.
- the feature identification module 406 a may then use such mappings 432 a - n to determine that future first inputs and second inputs map to the same common feature, without needing to apply natural language processing or machine learning techniques to do so. For example, when the system 700 of FIG. 7 receives a subsequent first input and second input (which may be the same as or differ from the previous first input and second input described above), the system 700 may apply the techniques disclosed above to look up the subsequent first input and the subsequent second input in the existing mappings 432 a - n (some of which may have been generated using the techniques described above in connection with FIGS.
- the method 600 of FIG. 6 and the system 700 of FIG. 7 may learn by updating the mappings 432 a - n and then apply the mappings 432 a - n to subsequent inputs.
- This is merely one example of a way in which embodiments of the present invention may learn automatically and store the knowledge resulting from such learning.
- the system 700 in response to determining that the first input and the second input both indicate the same feature as each other (e.g., in response to the mapping input 434 received from the user 436 indicating that the first input and the second input both indicate the same feature), the system 700 may apply any of a variety of machine learning techniques to learn from this determination. In this way, operation 614 in FIG.
- the feature identification module 406 a and/or other component of the system 700 may include a machine learning engine that receives the mapping input 434 and data representing the corresponding feature from the feature set data 404 , and automatically applies machine learning techniques to the mapping input 434 and corresponding feature data to learn an association (e.g., mapping) between the mapping input 434 and corresponding feature data. More generally, such machine learning techniques may be used to learn: (1) a first association between the first input and the feature F; and (2) a second association between the second input and the feature F.
- the system 700 may store data representing both such associations, such as by storing first data representing the first association and storing distinct second data representing the second association, or by storing data representing a three-way association among the first input, the second input, and the feature F.
- the system 700 may apply such learning to subsequent first and second inputs to determine automatically that such inputs indicate particular features.
- machine learning techniques that may be used in the manner described above include decision tree learning, association rule learning, artificial neural networks, deep learning, inductive logic programming, support vector machines, clustering, Bayesian networks, reinforcement learning, representation learning, similarity and metric learning, sparse dictionary learning, and genetic algorithms.
- machine learning software examples include dlib, ELKI, Encog, GNU Script Octave, H2O, Mahout, Mallet, mlpy, MLPACK, MOA, ND4J, NuPIC, OpenCV, OpenNN, Orange, R, scikit-learn, Shogun, TensorFlow, Torch, Spark, Yooreka, Weka, KNIME, RapidMinder, Angoss, Databricks, Google Prediction API, IBM SPSS Modeler, KXEN Modeler, LIONsolver, Mathematica, MATLAB, Microsoft Azure Machine Learning, Neural Designer, NeuroSolutions, Oracle Data Mining, RCASE, SAS Enterprise Miner, and STATISTICA Data Miner.
- the method 500 repeats the operations within the loop initiated in operation 504 ( FIG. 5 , operation 514 ), and repeats the operations within the loop initiated in operation 506 ( FIG. 5 , operation 516 ).
- the feature data 408 includes frequency counts for all of the features of all of the objects in the sample set. It should be appreciated that, alternatively, the method 500 may produce frequency counts for fewer than all features in the feature set, for one or more objects in the sample set. Similarly, it should be appreciated that, alternatively, the method 500 may produce frequency counts for fewer than all objects in the sample set.
- the method 500 may repeat one or more additional times (as illustrated by path 517 in FIG. 5 ).
- the method 500 may be performed once by each of a plurality of feature identification modules in the system (e.g., feature identification modules 406 a , 406 b , and 406 c ).
- the feature data 408 may be initialized only once (in operation 502 ), so that repeated performance of operations 504 - 516 causes frequency data produced by multiple feature identification modules to be combined (e.g., summed) with each other.
- the system 400 also includes an obscure feature identification module 418 , which may identify features of objects in the sample set having a particularly high frequency and/or features of objects in the sample set having a particularly low frequency, based on the feature count data 412 and/or the feature output 416 , thereby generating obscure feature data 420 , which indicates which features of the objects of the sample set have a particularly high frequency (i.e., features which satisfy a high frequency criterion) and/or which features of the objects in the sample set have a particularly low frequency (i.e., features which satisfy a low frequency criterion) ( FIG. 5 , operation 518 ).
- an obscure feature identification module 418 may identify features of objects in the sample set having a particularly high frequency and/or features of objects in the sample set having a particularly low frequency, based on the feature count data 412 and/or the feature output 416 , thereby generating obscure feature data 420 , which indicates which features of the objects of the sample set have a particularly high frequency (i.e., features
- the obscure feature identification module 418 may include: (1) one or more computers; (2) one or more humans; or (3) any combination of (1) and (2).
- the obscure feature identification module 418 may include a computer that automatically analyzes some or all of the feature count data 412 to produce some or all of the obscure feature data 420 based on some or all of the feature count data 412 .
- the obscure feature identification module 418 may include a human who manually analyzes some or all of the feature count data 412 to produce some or all of the obscure feature data 420 based on some or all of the feature count data 412 .
- the system 400 may include multiple obscure feature identification modules, which may in combination produce the obscure feature data 420 .
- Each of such multiple obscure feature identification modules may include: (1) one or more computers; (2) one or more humans; or (3) any combination of (1) and (2).
- the obscure feature identification module 418 may produce the obscure feature data 420 in any of a variety of ways. For example, the obscure feature identification module 418 may determine, for each of one or more features in the feature set, whether the feature count data 412 indicates that the feature has a particularly low frequency (i.e., that the feature satisfies a low frequency criterion), such as by determining whether the frequency count for that feature is less than some predetermined maximum value (e.g., 3, 2, or 1). As a particular example, the obscure feature identification module 418 may determine whether the frequency count of the feature is equal to zero.
- some predetermined maximum value e.g., 3, 2, or 1).
- the obscure feature identification module 418 may determine whether the frequency count of the feature is in the lowest X percentile of the frequency count data 412 , where X may be any value, such as 1, 2, 5, 10, or 20. If the obscure feature identification module 418 determines that the frequency count for a feature is particularly low, then the obscure feature identification module 418 may store an indication, in the obscure feature data 420 , that the feature has a particularly low frequency (i.e., is an obscure feature).
- the obscure feature identification module 418 may determine, for each of one or more features in the feature set, whether the feature count data 412 indicates that the feature has a particularly high frequency (i.e., that the feature satisfies a high frequency criterion), such as by determining whether the frequency count for that feature is greater than some predetermined minimum value (e.g., 3, 2, or 1). As another example, the obscure feature identification module 418 may determine whether the frequency count of the feature is in the highest X percentile of the frequency count data 412 , where X may be any value, such as 1, 2, 5, 10, or 20.
- the obscure feature identification module 418 may store an indication, in the obscure feature data 420 , that the feature has a particularly high frequency, or that the feature does not have a particularly low frequency (i.e., is not an obscure feature).
- the human(s) may make the determination in operation 518 of FIG. 5 by manually viewing the feature output (e.g., the bar chart of FIG. 1 ) and manually determining whether certain features have particularly low frequencies (e.g., frequencies of zero).
- the feature output e.g., the bar chart of FIG. 1
- the human(s) may make the determination in operation 518 of FIG. 5 by manually viewing the feature output (e.g., the bar chart of FIG. 1 ) and manually determining whether certain features have particularly low frequencies (e.g., frequencies of zero).
- the system 400 may include an obscure feature output module 422 , which may produce obscure feature output 424 based on the obscure feature data 420 ( FIG. 5 , operation 520 ).
- the obscure feature output 424 represents the obscure feature data 420 .
- the obscure feature output module 422 may produce the obscure feature output 424 in any of a variety of ways.
- the obscure feature output module 422 may produce the obscure feature output 424 in the form of a chart, such as a bar chart, a pie chart, or other chart representing the frequency counts in the obscure feature data 420 .
- the obscure features identified by the obscure feature data 420 may then be used to develop new instances of objects represented by the objects in the sample set, by developing new instances of objects having the obscure features represented by the obscure feature data 420 .
- Such development may, for example, be performed manually by humans after observing output representing the obscure feature data 420 , and then developing new instances of objects having features that are identified as obscure features by the obscure feature data 420 .
- Embodiments of the present invention may assist in this process by, for example, automatically producing the obscure feature output 424 in a form which emphasizes the features identified as obscure features by the obscure feature data 420 .
- the obscure feature output 424 may be generated by modifying the feature output 416 (e.g., the bar chart of FIG.
- the obscure feature output 424 may, for example, include output representing the obscure feature data 420 and not include output representing features not represented by the obscure feature data 420 , so that the obscure feature output 424 presents to the user only representations of obscure features in the sample set and no other (non-obscure) features in the sample set.
- the system 400 may provide such modified output to users of the system 400 to make it easier for such users to quickly and easily understand which features in the feature set are infrequently or never observed in the objects in the sample set.
- the feature identification modules 406 a - c may include any combination of humans and computers. More generally, various aspects of the system 400 may be implemented using computers, humans, or a combination thereof. For example:
- the feature output 416 may provide a panoramic view of the possible types of features, and their relative observed frequencies, in more of more objects in a class of objects. Such a panoramic view enables innovators to see the obscure feature types available for new designs as well as the feature types that previous solutions have been built upon.
- the obscure feature output 424 may emphasize obscure features in the sample set to the user, thereby enabling the user to quickly and easily identify obscure features in the sample set.
- the obscure feature output 424 takes the form of a chart which emphasizes obscure features in the sample set, the user may quickly identify obscure features with a quick glance at the chart, even if there is a large number of samples in the sample set and a large number of features in the feature set.
- Embodiments of the present invention may use any feature set containing any number and type of features in any combination.
- a feature set also referred to herein as a feature type taxonomy
- experiments that were conducted to develop the particular feature set will be described.
- FIG. 1 shows our results for a candle in the form of feature type spectrum (FTS), named as such because it gives a kind of spectral analysis to the features of a candle (McCaffrey and Spector, 2011).
- the y-axis of FIG. 11 represents the average number of times these subjects listed a feature of a particular type.
- the x-axis shows the 32 feature types presented by number.
- the feature type spectrum shown in FIG. 1 is an example of the feature output 416 in the system 400 of FIG. 4 .
- the frequencies illustrated by FIG. 1 are examples of the feature count data 412 in the system 400 of FIG. 4 .
- FIG. 1 shows a clear pattern of underexplored and ignored feature types that could become the basis for innovation.
- the low bars (representing low frequencies, e.g., low values in the feature count data 412 ) and non-existent bars (representing values of zero in the feature count data 412 ) of FIG. 1 point to the obscure feature types upon which to build new candle designs.
- FIG. 1 we were able to create ten new candle designs in two one-hour sessions.
- a feature type taxonomy disclosed herein is intended to be a taxonomy that generally applies to all physical objects and materials, in that it only contains types of features that can apply to all physical objects and materials.
- the particular feature type taxonomy disclosed herein is merely an example and does not constitute a limitation of the present invention. In practice, it may be used as a default or starting point, or it may be entirely replaced by other taxonomies.
- the particular example of a feature type taxonomy disclosed herein contains 32 categories of features, feature type taxonomies used in conjunction with embodiments of the present invention may contain any number of categories of features.
- the 32 feature types of the present example of a feature type taxonomy are segmented into two kinds: Physical Feature Types (14 feature types under this kind) and Use-Based Feature Types (18 features types under this kind).
- Physical Feature Types 14 feature types under this kind
- Use-Based Feature Types (18 features types under this kind).
- a speaker shows the first slide and narrates, “Here is a picture of something to sit on.”
- the second slide is shown. “Here is a picture of something to stand on to change a light bulb.”
- the third slide is shown. “Here is a picture of a homeplate for a whiffle ball game.”
- the fourth slide is shown. “Here is a picture of something to leverage under a doorknob to prevent someone from entering a room.”
- the fifth slide is shown. “Here is something to row with.” Turn the chair upside down, grab two legs, and start paddling water with the back of the chair pressing against the water.
- the sixth slide is shown. “Here is something that can provide shade for a short delicate plant that cannot tolerate direct sunlight.”
- the seventh slide is shown. “Here is something for shovelling a pile of leaves.” Grab a chair handle with one hand and a chair leg with another hand, and then start to shovel the leaves. There are many other slides, but we will stop here.
- Table 1 presents the 32 types of features that are included in one example of a feature type taxonomy according to embodiments of the present invention.
- the first 14 feature types are considered the physical features that have a certain independence from the object's use.
- the remaining 18 feature types are considered the use-based features that take on their values while the object is in use and change when the object is used in a different manner.
- the first column presents the name of the feature type.
- the second column gives a description of the feature type.
- the third column presents an example based on the common use of the plastic chair in FIG. 3 .
- focal entity Material Material make-up Legs are metal of focal entity or its parts Shape Overall shape of Legs are U-shaped focal entity or cylinders its parts Symmetry An important but Legs are often overlooked symmetrical in two characteristic of dimensions the shape of a focal entity Size Length, width, Legs are about 4 depth of focal feet long and have entity or its a diameter of 2 parts inches Color . . . of focal entity Legs are yellow or its parts Texture . . . of focal entity Legs are smooth or its parts Aroma . . .
- the seat of the (First of the Use- entity to chair relates to Based Features) environmental the seat of a entities during a person when the particular use of chair is being sat the focal entity.
- Environmental Environmental A chair is often Partners entities that the used with a table focal entity is or a desk. used with during a particular use Motor Relations How a human To sit in a chair physically requires a complex manipulates the motor movement focal entity or that involves its parts during a bending the knees particular use so that the seat of the person lands on the seat of the chair.
- the cause-effect weight is fairly sequence set off evenly distributed among the parts of across the chair's the focal entity seat.
- the weight as well as between stresses the the focal entity connecting points and its between chair seat environmental and the legs. entities (etc.) Place
- the typical Chairs often physical locations appear in that the focal kitchens, dining entity resides in rooms, offices, on during a decks, etc. particular use Occasion
- the typical Chairs are present contexts that a during a family focal entity meal or a cookout resides in during on one's deck.
- the typical type A chair is of motion engaged generally in by a focal motionless when it entity during a is being sat upon.
- Permanence/Transience How long the focal A chair is usually entity tends to designed to last last as it is used for many years.
- Superordinate The more general Based on its classification of designed use, the the focal entity superordinate of a based on its chair is typical use furniture.
- Subordinate More specific Based on its versions of the designed use, a focal entity based subordinate of a on its typical use chair is a rocking chair or a bench. Synonym (based on Other entities Other objects (not use) that can achieve subordinates) that the same use as can be sat on in a the focal entity pinch.
- Space The spatial Any spatial relations between relation between a the focal entity chair and other and the objects during its environmental designed use. entities during a Example: a chair particular use is pulled under a table so that the back of the chair is about 1.5 feet from the edge of the table Orientation The spatial In order to be sat orientation upon, the chair is required for the upright; that is, focal entity to the seat of the achieve its use (a chair is above the very important legs. sub-case of overall spatial relations) Side Effects Other effects A side effect of besides the sitting in a chair desired ones that is the pressure of are produced while the legs on the the focal entity floor. If used in is in use the same place on the floor, over time this pressure can create indentations on the floor. Sound The sound emitted A chair may creak by the focal when a heavy entity during a person sits on the particular use chair.
- feature type taxonomy shown in FIG. 2 is divided into two levels (types), this is merely an example and does not constitute a limitation of the present invention. More generally, feature type taxonomies used in conjunction with embodiments of the present invention may take any form.
- a feature type taxonomy may have a hierarchical (e.g., tree-shaped) form with any number of levels, branches, and nodes in any configuration.
- embodiments of the present invention may determine whether a particular object has a particular feature based on the feature data 408 that is output by the feature identification modules 406 a - c .
- the feature data 408 may include records of observations, memories, judgments, and other determinations (by computers and/or humans) of whether particular objects have particular features.
- Embodiments of the present invention may use such records of determinations as proxies for the actual features of the actual objects themselves.
- any reference herein to an object “having” a feature, parameter, or parameter value should be understood to refer to an indication (e.g., by the feature data 408 ) that the object has the feature, parameter, or parameter value (such as an indication resulting from a perception or conclusion by one or more of the feature identification modules 406 a - c that the object has the feature, parameter, or parameter value), whether or not the object actually has the feature, parameter, or parameter value.
- references herein to the “frequency” or “frequency of occurrence” of a feature, parameter, or parameter value with respect to a particular object should be understood to refer to the frequency with which the feature, parameter, or parameter value is indicated by the feature data 408 with respect to the particular object (e.g., the number of times the feature identification modules 406 a - c determine that the object has the feature, parameter, or parameter value).
- Certain observations of a particular object may result in a determination that the object has a particular feature, parameter, or parameter value, while other observations of the same object may not result in a determination that the object has the particular feature, parameter, or parameter value.
- a ceramic cup may be observed by three different people, two of whom may conclude that the cup has the material parameter value of “ceramic,” and one of whom may not conclude that the cup has the material parameter value of “ceramic.”
- features described herein as “use-based features” are statements about how an object may be used (e.g., the place of use or the occasion of use). For example, a ceramic cup often appears in restaurants, diners, and kitchens. These are examples of the ceramic cup's place of use. Examples of occasions of use for a ceramic cup may include: drinking a hot liquid with a meal and drinking coffee with breakfast. In these examples, the object (i.e., ceramic cup) does not inherently “have” the stated feature. Instead, the stated feature (e.g., the ceramic cup's place of use or occasion of use) describes circumstances commonly associated with the use of the object. Therefore, references herein to an object “having” a particular use-based feature, parameter, or parameter value refers to the fact that the object was observed or otherwise determined to have the particular use-based feature during the object's normal course of use.
- Any of the functions disclosed herein may be implemented using means for performing those functions. Such means include, but are not limited to, any of the components disclosed herein, such as the computer-related components described below.
- the techniques described above may be implemented, for example, in hardware, one or more computer programs tangibly stored on one or more computer-readable media, firmware, or any combination thereof.
- the techniques described above may be implemented in one or more computer programs executing on (or executable by) a programmable computer including any combination of any number of the following: a processor, a storage medium readable and/or writable by the processor (including, for example, volatile and non-volatile memory and/or storage elements), an input device, and an output device.
- Program code may be applied to input entered using the input device to perform the functions described and to generate output using the output device.
- Each computer program within the scope of the claims below may be implemented in any programming language, such as assembly language, machine language, a high-level procedural programming language, or an object-oriented programming language.
- the programming language may, for example, be a compiled or interpreted programming language.
- Embodiments of the present invention include features which are only possible and/or feasible to implement with the use of one or more computers, computer processors, and/or other elements of a computer system. Such features are either impossible or impractical to implement mentally and/or manually.
- embodiments of the present invention may automatically learn and store associations between inputs and features using computer-automated machine learning techniques. Such techniques are in fact performed by a computer, and are only capable of being performed by computers. As a result, such techniques are inherently computer-related.
- one result of applying such automated machine learning techniques is to enable embodiments of the present invention to apply such learning to future inputs to determine automatically that such inputs are associated with particular features.
- Such automated application of the results of machine learning is in fact performed by a computer, and can only be performed by a computer. As a result, such automated application of the results of machine learning is inherently computer-related.
- embodiments of the present invention provide inherently technical solutions to inherently technical problems.
- embodiments of the present invention provide inherently technical solutions to the inherently technical problem of how to use a computer to automatically learn that two or more inputs (e.g., text strings) are associated with the same feature as each other.
- This problem is inherently technical because it relates to the use of a computer to draw a conclusion about the meaning of inputs, even though a computer cannot understand meaning. Instead, if a computer is to conclude that two or more inputs are associated with the same feature as each other, it must be by using technical mechanisms to achieve the result of concluding accurately that such inputs are associated with the same feature as each other, but without understanding the meanings of such inputs.
- Embodiments of the present invention solve this inherently technical problem by applying computer-automated techniques, such as computer-automated machine learning techniques, to determine that two or more inputs are associated with the same feature as each other.
- embodiments of the present invention provide inherently technical solutions to the inherently technical problem of how to use sensors to perform sensing operations to generate data representing sensed properties of a physical object.
- This is an inherently technical problem because it relates to the use of machinery, namely sensors, to sense physical properties of physical objects automatically.
- Embodiments of the present invention solve this inherently technical problem by using sensors to perform sensing operations to generate data representing sensed properties of a physical object, and using a computer to map such data to features in a feature set automatically.
- One technical benefit of such solutions provided by embodiments of the present invention is that they enable the physical properties of a physical object to be identified more quickly and with less human effort (possibly no human effort) than by relying on human senses and input to identify the properties of the physical object.
- Each such computer program may be implemented in a computer program product tangibly embodied in a machine-readable storage device for execution by a computer processor.
- Method steps of the invention may be performed by one or more computer processors executing a program tangibly embodied on a computer-readable medium to perform functions of the invention by operating on input and generating output.
- Suitable processors include, by way of example, both general and special purpose microprocessors.
- the processor receives (reads) instructions and data from a memory (such as a read-only memory and/or a random access memory) and writes (stores) instructions and data to the memory.
- Storage devices suitable for tangibly embodying computer program instructions and data include, for example, all forms of non-volatile memory, such as semiconductor memory devices, including EPROM, EEPROM, and flash memory devices; magnetic disks such as internal hard disks and removable disks; magneto-optical disks; and CD-ROMs. Any of the foregoing may be supplemented by, or incorporated in, specially-designed ASICs (application-specific integrated circuits) or FPGAs (Field-Programmable Gate Arrays).
- a computer can generally also receive (read) programs and data from, and write (store) programs and data to, a non-transitory computer-readable storage medium such as an internal disk (not shown) or a removable disk.
- Any data disclosed herein may be implemented, for example, in one or more data structures tangibly stored on a non-transitory computer-readable medium. Embodiments of the invention may store such data in such data structure(s) and read such data from such data structure(s).
Abstract
Description
-
- location sensors (such as Global Positioning System (GPS) sensors, Bluetooth Low Energy Beacons, or Wi-Fi Positioning System (WPS) sensors), in which case the sample set
data 402 may include data representing one or more locations (e.g., one or more locations of the object O); - motion sensors, in which case the sample set
data 402 may include data representing one or more physical motions of the object O; - acoustic sensors (such as a geophone, hydrophone, or microphone), in which case the sample set
data 402 may include data representing one or more acoustic characteristics of the object O, such as data representing characteristics (e.g., pitch and/or amplitude) of sounds emitted by the object O; - chemical sensors (such as breathalyzers, carbon dioxide sensors, and oxygen sensors), in which case the sample set
data 402 may include data representing chemical characteristics (e.g., chemical composition and/or reaction rate) of one or more chemicals in, on, or emitted by the object O; - electric current, electric potential, magnetic, and radio sensors (such as current sensors, galvanometers, magnetometers, and voltage detectors), in which case the sample set
data 402 may include data representing one or more electrical characteristics of the object O, such as one or more of electric current, electrical potential, resistance, magnetic fields, conductivity, or radio waves emitted by or otherwise sensed from the object O; - radioactivity sensors, in which case the sample set
data 402 may include data representing one or more radioactivity characteristics of the object O, such as decay rate and/or intensity; - flow and fluid velocity sensors, such as air flow meters, anemometers, flow sensors, gas meters, mass flow sensors, and water meters, in which case the sample set
data 402 may data representing one or more flow and/or fluid velocity characteristics of the object O, such as flow and/or fluid velocity sensed in connection with the object O; - position, angle, displacement, distance, speed/velocity, momentum, vibration, and acceleration sensors, such as capacitive displacement sensors, capacitive sensing sensors, free fall sensors, gyroscopic sensors, impact sensors, inclinometers, integrated circuit piezoelectric sensors, liquid capacitive inclinometers, odometers, photoelectric sensors, piezocapacitive sensors, piezoelectric accelerometers, position sensors, tilt sensors, tachometers, and velocity receivers, in which case the sample set
data 402 may include data representing any combination of position, angle, displacement, distance, speed, and acceleration of the object O; - optical, light, imaging, and photon sensors, such as cameras, charge-coupled devices, CMOS sensors, colorimeters, contact image sensors, electro-optical sensors, infra-red sensors, kinetic inductance detectors, LED as light sensors, optical position sensors, photodetectors, photodiodes, phototransistors, photoelectric sensors, and photoresistors, in which case the sample set
data 402 may include data representing one or more sensed optical inputs from the object O (which may, for example, be stored in the form of images and/or video); - pressure sensors, such as barographs, barometers, piezometers, pressure gauges, and tactile sensors, in which case the sample set
data 402 may include data representing one or more pressure characteristics sensed from the object O; - force, density, level, tension, pressure, balance, friction, gravity, centrifugal force, centripetal force, and torque sensors, such as piezocapacitive pressure sensors, piezoelectric sensors, strain gauges, and torque sensors, in which case the sample set
data 402 may include data representing any force, density, or level input sensed from the object O; - thermal, heat, and temperature sensors, such as calorimeters, infrared thermometers, resistance temperature detectors, resistance thermometers, temperature gauges, thermistors, thermocouples, thermometers, and pyrometers, in which case the sample set
data 402 may include data representing any one or more thermal, heat, or temperature inputs sensed from the object O; - proximity and presence sensors, such as alarm sensors, Doppler radar sensors, motion detectors, proximity sensors, passive infrared sensors, touch switches, and wired gloves, in which case the sample set
data 402 may include data representing any one or more proximity or presence inputs sensed from the object O; and - durability sensors, in which case the sample set
data 402 may include data representing any one or more durability characteristics of the object O, such as the strength and/or toughness of the object O.
- location sensors (such as Global Positioning System (GPS) sensors, Bluetooth Low Energy Beacons, or Wi-Fi Positioning System (WPS) sensors), in which case the sample set
-
- if the first input is the text string “color”;
- if the second input is the text string “pigmentation”;
- if the feature identification module maps the first input to the feature of “color”;
- if the feature identification module maps the second input to the feature of “color”;
- then the
feature identification module 406 a may determine that the first feature is the same feature as the second feature; and - the
feature identification module 406 a may determine that the first input and the second input both indicate that the object O has the feature of “color.”
-
- a first mapping indicating that the first input maps to the common feature; and
- a second mapping indicating that the second input maps to the common feature.
-
- The sample set
data 402 may, for example, be stored as data in a non-transitory computer-readable medium and in a format that is readable by a computer. Additionally or alternatively, for example, the sample setdata 402 may be analyzable by humans without the aid of a computer. For example, the sample set may be or include the objects in the sample set themselves, or data representing the sample set in a format that may be analyzed by humans without the use of a computer, such as printed photographs of the objects in the sample set. - The feature set
data 404 may, for example, be stored as data in a non-transitory computer-readable medium and in a format that is readable by a computer. Additionally or alternatively, for example, the feature setdata 404 may be analyzable by humans without the aid of a computer. For example, the feature setdata 404 may be implemented as a list of descriptions of features in the feature set, written on paper. - The
feature data 408 may, for example, be stored as data in a non-transitory computer-readable medium and in a format that is readable by a computer. Additionally or alternatively, for example, thefeature data 408 may be created and analyzable by humans partially or entirely without the aid of a computer. For example, thefeature data 408 may be a description, written on paper or typed into a word processing document by human observers, of the presence/absence of features from the feature set in the objects in the sample set. - The
feature count data 412 may, for example, be stored as data in a non-transitory computer-readable medium and in a format that is readable by a computer. Additionally or alternatively, for example, thefeature count data 412 may be created and analyzable by humans partially or entirely without the aid of a computer. For example, thefeature count data 412 may be a description, written on paper or typed into a word processing document by human observers, of the count of the number of observations of each feature from the feature set in the objects in the sample set. As described above, the count of observations of a particular feature for a particular object may be the sum of the number of observations of that feature in that object across all of the feature identification modules (some or all of which may be humans). - The obscure feature data 420 may, for example, be stored as data in a non-transitory computer-readable medium and in a format that is readable by a computer. Additionally or alternatively, for example, the obscure feature data 420 may be created and analyzable by humans partially or entirely without the aid of a computer. For example, the obscure feature data 420 may be a description, written on paper or typed into a word processing document by human observers, of features of objects in the sample set having particular high and/or particularly low frequencies of observation.
- The
obscure feature output 424 may, for example, be stored as data in a non-transitory computer-readable medium and in a format that is readable by a computer. Additionally or alternatively, for example, theobscure feature output 424 may be created and analyzable by humans partially or entirely without the aid of a computer.
- The sample set
TABLE 1 |
Example Feature Type Taxonomy |
Example (based on | ||
plastic chair in | ||
Name | Description | FIG. 3) |
Parts | Identifiable | Legs |
(First of the | components of | |
Physical Features) | focal entity | |
Material | Material make-up | Legs are metal |
of focal entity or | ||
its parts | ||
Shape | Overall shape of | Legs are U-shaped |
focal entity or | cylinders | |
its parts | ||
Symmetry | An important but | Legs are |
often overlooked | symmetrical in two | |
characteristic of | dimensions | |
the shape of a | ||
focal entity | ||
Size | Length, width, | Legs are about 4 |
depth of focal | feet long and have | |
entity or its | a diameter of 2 | |
parts | inches | |
Color | . . . of focal entity | Legs are yellow |
or its parts | ||
Texture | . . . of focal entity | Legs are smooth |
or its parts | ||
Aroma | . . . of focal entity | No aroma for legs |
or its parts | ||
Number | Number of | 2 legs (because of |
components of a | the U-shape) | |
certain kind of | ||
the focal entity | ||
of its parts | ||
Mass | . . . of focal entity | The mass of the |
or its parts | chair. | |
Weight | . . . of focal entity | A U-shaped leg |
or its parts | weighs about 1 | |
pound | ||
State of Matter | (Solid, liquid, | Legs are solid |
gas, plasma) of | ||
focal entity or | ||
its parts | ||
Connectivity among | Physical | The legs are |
Parts | connection among | connected to the |
components of the | seat | |
focal entity. This | ||
feature is based | ||
on the chair when | ||
it is not being | ||
used. An inert | ||
chair possesses | ||
this feature of | ||
its parts being | ||
connected in some | ||
way. | ||
Spatial Relations | Distance and | The bottoms of all |
among Parts | direction of one | four legs form a |
component to | plane. | |
another of the | ||
focal entity. | ||
Again, this | ||
feature is based | ||
on the chair when | ||
it is not being | ||
used. An inert | ||
chair possesses | ||
this feature of | ||
their being | ||
spatial relations | ||
among the parts. | ||
External Relations | Relations of focal | The seat of the |
(First of the Use- | entity to | chair relates to |
Based Features) | environmental | the seat of a |
entities during a | person when the | |
particular use of | chair is being sat | |
the focal entity. | upon by the person | |
Environmental | Environmental | A chair is often |
Partners | entities that the | used with a table |
focal entity is | or a desk. | |
used with during a | ||
particular use | ||
Motor Relations | How a human | To sit in a chair |
physically | requires a complex | |
manipulates the | motor movement | |
focal entity or | that involves | |
its parts during a | bending the knees | |
particular use | so that the seat | |
of the person | ||
lands on the seat | ||
of the chair. | ||
Causal Relations | During a | When a person sits |
particular use, | on a chair, the | |
the cause-effect | weight is fairly | |
sequence set off | evenly distributed | |
among the parts of | across the chair's | |
the focal entity | seat. The weight | |
as well as between | stresses the | |
the focal entity | connecting points | |
and its | between chair seat | |
environmental | and the legs. | |
entities | (etc.) | |
Place | The typical | Chairs often |
physical locations | appear in | |
that the focal | kitchens, dining | |
entity resides in | rooms, offices, on | |
during a | decks, etc. | |
particular use | ||
Occasion | The typical | Chairs are present |
contexts that a | during a family | |
focal entity | meal or a cookout | |
resides in during | on one's deck. | |
a particular use | ||
Energy/Forces | During a | Because the chair |
particular use, | is plastic, static | |
the types of | electricity often | |
energy and forces | builds up between | |
in play both | the chair surface | |
within the focal | and the clothes of | |
entity as well as | the person using | |
within and among | the chair. | |
the environmental | ||
entities | ||
Perspective | The typical | A person of views |
physical viewing | the chair from a | |
point that a human | vantage point of | |
takes with respect | several feet above | |
to the focal | the chair and | |
entity during a | several to many | |
particular use. | feet away from the | |
chair. The typical | ||
perspective shapes | ||
what parts of the | ||
chair people tend | ||
to notice and | ||
which parts they | ||
overlook. | ||
Time | The typical time- | An occasion of |
frame | sitting commonly | |
(milliseconds, | lasts between | |
hours) that a | several minutes to | |
focal entity | a couple of hours. | |
occupies during a | ||
particular use | ||
Motion | The typical type | A chair is |
of motion engaged | generally | |
in by a focal | motionless when it | |
entity during a | is being sat upon. | |
particular use | ||
Permanence/Transience | How long the focal | A chair is usually |
entity tends to | designed to last | |
last as it is used | for many years. | |
Superordinate | The more general | Based on its |
classification of | designed use, the | |
the focal entity | superordinate of a | |
based on its | chair is | |
typical use | furniture. | |
Subordinate | More specific | Based on its |
versions of the | designed use, a | |
focal entity based | subordinate of a | |
on its typical use | chair is a rocking | |
chair or a bench. | ||
Synonym (based on | Other entities | Other objects (not |
use) | that can achieve | subordinates) that |
the same use as | can be sat on in a | |
the focal entity | pinch. Examples: a | |
large flat rock, a | ||
kitchen counter, a | ||
coffee table. | ||
Space | The spatial | Any spatial |
relations between | relation between a | |
the focal entity | chair and other | |
and the | objects during its | |
environmental | designed use. | |
entities during a | Example: a chair | |
particular use | is pulled under a | |
table so that the | ||
back of the chair | ||
is about 1.5 feet | ||
from the edge of | ||
the table | ||
Orientation | The spatial | In order to be sat |
orientation | upon, the chair is | |
required for the | upright; that is, | |
focal entity to | the seat of the | |
achieve its use (a | chair is above the | |
very important | legs. | |
sub-case of | ||
overall spatial | ||
relations) | ||
Side Effects | Other effects | A side effect of |
besides the | sitting in a chair | |
desired ones that | is the pressure of | |
are produced while | the legs on the | |
the focal entity | floor. If used in | |
is in use | the same place on | |
the floor, over | ||
time this pressure | ||
can create | ||
indentations on | ||
the floor. | ||
Sound | The sound emitted | A chair may creak |
by the focal | when a heavy | |
entity during a | person sits on the | |
particular use | chair. | |
Claims (44)
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US15/142,099 US9646266B2 (en) | 2012-10-22 | 2016-04-29 | Feature type spectrum technique |
US15/467,988 US20170193339A1 (en) | 2012-10-22 | 2017-03-23 | Feature Type Spectrum Technique |
US15/716,843 US20180114102A1 (en) | 2012-10-22 | 2017-09-27 | Feature Type Spectrum Technique |
Applications Claiming Priority (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US201261716681P | 2012-10-22 | 2012-10-22 | |
US14/059,578 US20140125676A1 (en) | 2012-10-22 | 2013-10-22 | Feature Type Spectrum Technique |
US15/142,099 US9646266B2 (en) | 2012-10-22 | 2016-04-29 | Feature type spectrum technique |
Related Parent Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US14/059,578 Continuation-In-Part US20140125676A1 (en) | 2012-10-22 | 2013-10-22 | Feature Type Spectrum Technique |
Related Child Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US15/467,988 Continuation US20170193339A1 (en) | 2012-10-22 | 2017-03-23 | Feature Type Spectrum Technique |
Publications (2)
Publication Number | Publication Date |
---|---|
US20160247091A1 US20160247091A1 (en) | 2016-08-25 |
US9646266B2 true US9646266B2 (en) | 2017-05-09 |
Family
ID=56690490
Family Applications (3)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US15/142,099 Active US9646266B2 (en) | 2012-10-22 | 2016-04-29 | Feature type spectrum technique |
US15/467,988 Abandoned US20170193339A1 (en) | 2012-10-22 | 2017-03-23 | Feature Type Spectrum Technique |
US15/716,843 Abandoned US20180114102A1 (en) | 2012-10-22 | 2017-09-27 | Feature Type Spectrum Technique |
Family Applications After (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US15/467,988 Abandoned US20170193339A1 (en) | 2012-10-22 | 2017-03-23 | Feature Type Spectrum Technique |
US15/716,843 Abandoned US20180114102A1 (en) | 2012-10-22 | 2017-09-27 | Feature Type Spectrum Technique |
Country Status (1)
Country | Link |
---|---|
US (3) | US9646266B2 (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10946592B2 (en) * | 2016-09-11 | 2021-03-16 | Impossible Objects, Inc. | Resistive heating-compression method and apparatus for composite-based additive manufacturing |
CN106953862B (en) * | 2017-03-23 | 2020-09-25 | 全球能源互联网研究院有限公司 | Sensing method and device for network security situation and sensing model training method and device |
CN111414750B (en) * | 2020-03-18 | 2023-08-18 | 北京百度网讯科技有限公司 | Synonym distinguishing method, device, equipment and storage medium |
US11380181B2 (en) * | 2020-12-09 | 2022-07-05 | MS Technologies | Doppler radar system with machine learning applications for fall prediction and detection |
Citations (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070282811A1 (en) | 2006-01-03 | 2007-12-06 | Musgrove Timothy A | Search system with query refinement and search method |
US20080104071A1 (en) | 2006-10-31 | 2008-05-01 | Execue, Inc. | System and method for converting a natural language query into a logical query |
US20080195570A1 (en) | 2005-03-31 | 2008-08-14 | Koninklijke Philips Electronics, N.V. | System and Method for Collecting Evidence Pertaining to Relationships Between Biomolecules and Diseases |
US20090070322A1 (en) | 2007-08-31 | 2009-03-12 | Powerset, Inc. | Browsing knowledge on the basis of semantic relations |
US20090100053A1 (en) | 2007-10-10 | 2009-04-16 | Bbn Technologies, Corp. | Semantic matching using predicate-argument structure |
US20090119095A1 (en) | 2007-11-05 | 2009-05-07 | Enhanced Medical Decisions. Inc. | Machine Learning Systems and Methods for Improved Natural Language Processing |
US20100070517A1 (en) | 2008-09-17 | 2010-03-18 | Oracle International Corporation | System and Method for Semantic Search in an Enterprise Application |
US20110022635A1 (en) | 2009-07-26 | 2011-01-27 | Moris Michael | Method and System to Formulate Queries With Minivisuals |
US20110161070A1 (en) | 2009-12-31 | 2011-06-30 | International Business Machines Corporation | Pre-highlighting text in a semantic highlighting system |
US20110173174A1 (en) | 2010-01-13 | 2011-07-14 | Flitcroft Investments Ltd | Linguistically enhanced search engine and meta-search engine |
US20110295823A1 (en) | 2010-05-25 | 2011-12-01 | Nokia Corporation | Method and apparatus for modeling relations among data items |
US20110301941A1 (en) | 2009-03-20 | 2011-12-08 | Syl Research Limited | Natural language processing method and system |
US20120254143A1 (en) | 2011-03-31 | 2012-10-04 | Infosys Technologies Ltd. | Natural language querying with cascaded conditional random fields |
US8602793B1 (en) | 2006-07-11 | 2013-12-10 | Erwin Ernest Sniedzins | Real time learning and self improvement educational system and method |
US20140059078A1 (en) | 2012-08-27 | 2014-02-27 | Microsoft Corporation | Semantic query language |
US8700620B1 (en) | 2010-04-27 | 2014-04-15 | Jeremy Lieberman | Artificial intelligence method and apparatus |
US8838659B2 (en) | 2007-10-04 | 2014-09-16 | Amazon Technologies, Inc. | Enhanced knowledge repository |
US8880547B1 (en) | 2009-07-21 | 2014-11-04 | Amazon Technologies, Inc. | Implementation of a search for items |
US8909516B2 (en) | 2011-10-27 | 2014-12-09 | Microsoft Corporation | Functionality for normalizing linguistic items |
US8924198B2 (en) | 2010-07-26 | 2014-12-30 | Radiant Logic, Inc. | Searching and browsing of contextual information |
US8943094B2 (en) | 2009-09-22 | 2015-01-27 | Next It Corporation | Apparatus, system, and method for natural language processing |
-
2016
- 2016-04-29 US US15/142,099 patent/US9646266B2/en active Active
-
2017
- 2017-03-23 US US15/467,988 patent/US20170193339A1/en not_active Abandoned
- 2017-09-27 US US15/716,843 patent/US20180114102A1/en not_active Abandoned
Patent Citations (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080195570A1 (en) | 2005-03-31 | 2008-08-14 | Koninklijke Philips Electronics, N.V. | System and Method for Collecting Evidence Pertaining to Relationships Between Biomolecules and Diseases |
US20070282811A1 (en) | 2006-01-03 | 2007-12-06 | Musgrove Timothy A | Search system with query refinement and search method |
US8602793B1 (en) | 2006-07-11 | 2013-12-10 | Erwin Ernest Sniedzins | Real time learning and self improvement educational system and method |
US20080104071A1 (en) | 2006-10-31 | 2008-05-01 | Execue, Inc. | System and method for converting a natural language query into a logical query |
US20090070322A1 (en) | 2007-08-31 | 2009-03-12 | Powerset, Inc. | Browsing knowledge on the basis of semantic relations |
US8838659B2 (en) | 2007-10-04 | 2014-09-16 | Amazon Technologies, Inc. | Enhanced knowledge repository |
US20090100053A1 (en) | 2007-10-10 | 2009-04-16 | Bbn Technologies, Corp. | Semantic matching using predicate-argument structure |
US20090119095A1 (en) | 2007-11-05 | 2009-05-07 | Enhanced Medical Decisions. Inc. | Machine Learning Systems and Methods for Improved Natural Language Processing |
US20100070517A1 (en) | 2008-09-17 | 2010-03-18 | Oracle International Corporation | System and Method for Semantic Search in an Enterprise Application |
US20110301941A1 (en) | 2009-03-20 | 2011-12-08 | Syl Research Limited | Natural language processing method and system |
US8880547B1 (en) | 2009-07-21 | 2014-11-04 | Amazon Technologies, Inc. | Implementation of a search for items |
US20110022635A1 (en) | 2009-07-26 | 2011-01-27 | Moris Michael | Method and System to Formulate Queries With Minivisuals |
US8943094B2 (en) | 2009-09-22 | 2015-01-27 | Next It Corporation | Apparatus, system, and method for natural language processing |
US20110161070A1 (en) | 2009-12-31 | 2011-06-30 | International Business Machines Corporation | Pre-highlighting text in a semantic highlighting system |
US20110173174A1 (en) | 2010-01-13 | 2011-07-14 | Flitcroft Investments Ltd | Linguistically enhanced search engine and meta-search engine |
US8700620B1 (en) | 2010-04-27 | 2014-04-15 | Jeremy Lieberman | Artificial intelligence method and apparatus |
US20110295823A1 (en) | 2010-05-25 | 2011-12-01 | Nokia Corporation | Method and apparatus for modeling relations among data items |
US8924198B2 (en) | 2010-07-26 | 2014-12-30 | Radiant Logic, Inc. | Searching and browsing of contextual information |
US20120254143A1 (en) | 2011-03-31 | 2012-10-04 | Infosys Technologies Ltd. | Natural language querying with cascaded conditional random fields |
US8909516B2 (en) | 2011-10-27 | 2014-12-09 | Microsoft Corporation | Functionality for normalizing linguistic items |
US20140059078A1 (en) | 2012-08-27 | 2014-02-27 | Microsoft Corporation | Semantic query language |
Also Published As
Publication number | Publication date |
---|---|
US20160247091A1 (en) | 2016-08-25 |
US20180114102A1 (en) | 2018-04-26 |
US20170193339A1 (en) | 2017-07-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20180114102A1 (en) | Feature Type Spectrum Technique | |
Hearst et al. | An evaluation of semantically grouped word cloud designs | |
Perry et al. | Assessing model fit: Caveats and recommendations for confirmatory factor analysis and exploratory structural equation modeling | |
Feinberg | A design perspective on data | |
Ding et al. | Approaches to data analysis of multiple-choice questions | |
Martin | Revisiting field: Specialized knowledge in secondary school science and humanities discourse | |
Kelley et al. | Using particle tracking to measure flow instabilities in an undergraduate laboratory experiment | |
Częstochowska et al. | On the context-free ambiguity of emoji | |
Yarom et al. | What you see is what you read? improving text-image alignment evaluation | |
Hullman et al. | Improving comprehension of measurements using concrete re-expression strategies | |
Hernández‐Carrasco et al. | Seasonal and regional characterization of horizontal stirring in the global ocean | |
Mulligan et al. | Promoting early mathematical structural development through an integrated assessment and pedagogical program | |
Frankenhuis et al. | Strategic ambiguity in the social sciences | |
Willey et al. | Visual field dependence as a navigational strategy | |
Vatavu | Beyond features for recognition: human-readable measures to understand users’ whole-body gesture performance | |
Spoladore et al. | Semantic and dweller-based decision support system for the reconfiguration of domestic environments: RecAAL | |
Byrne et al. | Figurative frames: A critical vocabulary for images in information visualization | |
Klavan | Pitting corpus-based classification models against each other: A case study for predicting constructional choice in written Estonian | |
Chamecki et al. | A TKE‐based framework for studying disturbed atmospheric surface layer flows and application to vertical velocity variance over canopies | |
Haber et al. | Word sense distance in human similarity judgements and contextualised word embeddings | |
JP2019128634A (en) | Optimization apparatus and optimization method for hyper parameter | |
Wang et al. | Association rules-based multivariate analysis and visualization of spatiotemporal climate data | |
Kane et al. | Modeling group differences in OLS and orthogonal regression: Implications for differential validity studies | |
US20130251264A1 (en) | Analysis of hand-drawn input groups | |
Cousyn et al. | Web-based objects detection to discover key objects in human activities |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AS | Assignment |
Owner name: UNIVERSITY OF MASSACHUSETTS, MASSACHUSETTS Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:MCCAFFREY, ANTHONY;REEL/FRAME:039560/0143 Effective date: 20130717 |
|
AS | Assignment |
Owner name: INNOVATION ACCELERATOR, INC., MASSACHUSETTS Free format text: LICENSE;ASSIGNOR:UNIVERSITY OF MASSACHUSETTS;REEL/FRAME:039805/0088 Effective date: 20130717 |
|
STCF | Information on status: patent grant |
Free format text: PATENTED CASE |
|
AS | Assignment |
Owner name: NATIONAL SCIENCE FOUNDATION, VIRGINIA Free format text: CONFIRMATORY LICENSE;ASSIGNOR:UNIVERSITY OF MASSACHUSETTS, AMHERST;REEL/FRAME:052044/0892 Effective date: 20190115 |
|
AS | Assignment |
Owner name: NATIONAL SCIENCE FOUNDATION, VIRGINIA Free format text: CONFIRMATORY LICENSE;ASSIGNOR:UNIVERSITY OF MASSACHUSETTS, AMHERST;REEL/FRAME:054279/0595 Effective date: 20200129 |
|
MAFP | Maintenance fee payment |
Free format text: PAYMENT OF MAINTENANCE FEE, 4TH YR, SMALL ENTITY (ORIGINAL EVENT CODE: M2551); ENTITY STATUS OF PATENT OWNER: SMALL ENTITY Year of fee payment: 4 |
|
AS | Assignment |
Owner name: NATIONAL SCIENCE FOUNDATION, VIRGINIA Free format text: CONFIRMATORY LICENSE;ASSIGNOR:UNIVERSITY OF MASSACHUSETTS AMHERST;REEL/FRAME:062205/0141 Effective date: 20200129 |